BlockBeats News, July 15th, according to CNBC, Apple is in preliminary talks with the AI startup PrismML to evaluate its AI model compression technology, aiming to enable large AI models to run directly on the iPhone without relying on cloud computing.

PrismML claims that it can reduce model memory usage by 10 to 15 times, allowing a model with 27 billion parameters to run on iPhone 15 and above devices. This results in a 6 to 8 times improvement in response speed and a 3 to 6 times reduction in energy consumption. If the technology matures, Apple could potentially move more Siri and generative AI functions to run locally, further enhancing response speed, privacy protection capabilities, offline user experience, and reducing cloud computing costs.

However, industry experts point out that this technology still needs to undergo large-scale real-world validation, and its impact on battery life, stability, and long-term performance remains to be seen.